Prevalence of Overweight and At Risk of Overweight in Fourth-Grade Children Across Five School-Based Studies Conducted During Four School Years

Please note that this study was published before the implementation of Healthy, Hunger-Free Kids Act of 2010, which went into effect during the 2012-13 school year, and its provision for Smart Snacks Nutrition Standards for Competitive Food in Schools, implemented during the 2014-15 school year. As such, certain research may not be relevant today.

The health consequences of overweight and obesity are public health issues that are among the most burdensome faced in the United States (Office of the Surgeon General, 2001). Overweight youth are at increased risk for adverse levels of several cardiovascular disease risk factors (Freedman et al., 1999); in addition, type-two diabetes, early maturation, and orthopedic problems are exhibited more frequently in overweight youth (Office of the Surgeon General, 2001). Furthermore, youth with high body mass index (BMI) for age percentiles have high risks of being overweight or obese in adulthood (Guo et al., 2002). The use of BMI (kg/m2) is recommended to identify youth who are overweight or at risk of becoming overweight (U.S. Centers for Disease Control and Prevention [CDC], 2006a). Based on the CDC’s gender-specific BMI-for-age growth charts for ages two to 20 years, youth with a BMI-for-age below the fifth percentile are categorized as underweight, youth at or above the fifth percentile and up to the 85th percentile are categorized as healthy weight, youth at or above the 85th percentile and up to the 95th percentile are categorized as at risk of overweight, and youth at or above the 95th percentile are categorized as overweight (CDC, 2006b).

National survey data have shown an increase in the prevalence of overweight among youth in the United States (CDC, 2006a). In the National Health and Nutrition Examination Survey (NHANES) III conducted 1988-94, rates of overweight were 11% for all youth ages six to 19 years; rates rose to 16% in NHANES 1999-2002 (CDC, 2006a). Therefore, over a period of approximately ten years, there was a 5% increase in the rate of overweight among youth.

Although there have been continuing increases in overweight among youth, the overall prevalence showed no change between NHANES 1999-2000 and NHANES 2001-02 (Hedley et al., 2004). Studies have found differences in overweight prevalence among youth by ethnicity (Hedley et al., 2004; Kimm et al., 2002; Melnik et al., 1998; Ogden et al., 2002), gender (Hedley et al., 2004; Melnik et al., 1998), and socioeconomic status (SES) (Moore et al., 2002; Wolfe et al., 1994).

This article discusses differences in BMI-for-age percentiles by ethnicity (Black, White), gender, and time (season/year), and prevalence of overweight and at risk of overweight, in Fourth-Grade children (approximately nine- to ten-years-old) across five school-based studies conducted during four school years between Spring 2000 (Study One) and Fall 2002 (Study Five). These studies concerned either the accuracy of children’s dietary recalls or children’s social desirability. Thus, weight and height measurements were secondary rather than primary aims.

Results concerning the accuracy of children’s dietary recalls (Baxter et al., 2004b; Baxter et al., 2002; Baxter, Thompson, Litaker, et al., 2003; Baxter, Thompson, Smith, et al., 2003) and children’s social desirability (Baxter et al., 2004a) are reported elsewhere.

Methodology

 An Institutional Review Board approved each study, and child assent and parental consent were obtained in writing prior to data collection. Children were recruited from all Fourth-Grade classes from a total of 14 public elementary schools in Richmond County, GA. The schools were selected based on high participation in school meals (breakfast, lunch), because four of the five studies used observation of school meals to validate these parts of children’s dietary recalls.

Table 1 provides details by study, including the number of schools; eligibility for free or

reduced-price school meals; agreed participation rates; number of children measured by ethnicity and gender; and month, year, and time of day of measurements. Information concerning children’s ethnicity, gender, and date of birth was obtained from schools.

Weight and height were measured by research staff in a private location at school. Children removed their shoes and heavy jackets prior to being measured. Weights were measured using digital scales (calibrated daily) according to a written protocol and recorded to the nearest tenth of a pound and heights were measured using portable stadiometers according to standardized procedures and recorded to the nearest eighth of an inch (Lohman et al., 1988; U.S. Department of Health and Human Services, n.d. Weight and height were measured once per child for Study One and twice per child for Studies Two through Five. If the two weight or height measurements per child were not within a tenth of a pound or a quarter of an inch, respectively, then a third weight or height was measured. If three weight and height measurements were obtained per child, then the average of the closest two was used for the child’s weight and/or height, respectively.

Because measurements were obtained by more than one research team member for each study, inter-rater reliability was assessed. For training purposes, research team members measured adults at the research office. While measuring children in schools, inter-rater reliability was assessed across pairs of research team members; on a daily basis, approximately 10% of all children measured were randomly selected and measured by both research team members.

Results from inter-rater reliability showed that for all five studies, the intraclass correlation reliability was greater than 0.99 for weight and for height. These inter-rater reliability results indicate that there was good agreement on weight and height measurements for the randomly selected children.

The child’s age at the time of measurement was calculated by subtracting the date of birth from the date of measurement. The CDC’s gender-specific BMI-for-age growth charts for youth aged two to 20 years were used to determine each child’s BMI-for-age percentile (Kuczmarski et al., 2002). The primary dependent variable was the child’s BMI-for-age percentile. Analysis of covariance was used to compare BMI-for-age percentiles by ethnicity (Black, White) and gender across study times. Due to differences in the timing of data collection for the five studies, time codes were assigned to reflect the season and year of measurement of the four school years (1 = Spring 2000; 2 = Spring 2001; 3 = Spring 2002; and 3.5 = Fall 2002). Percentile data were rank- transformed due to asymmetry of the sample distributions. Children were categorized as underweight, healthy weight, at risk of overweight, or overweight according to their BMI-for-age percentile.

Percentages of children in the overweight category and the category for at risk of overweight were compared by ethnicity and gender using Pearson’s chi-square statistic and across times using the Mantel-Haenszel chi-square test for trend. To maintain an overall alpha level of 0.05, a Bonferroni adjustment for multiple comparisons was used. Children of other ethnic groups (n=73) were excluded from analysis due to the small number. Because eight children who repeated fourth grade had participated in more than one of the five studies, only measurements from the first study, in which an individual child participated, were used for data analysis.

Analyses were conducted using SAS®. Child was the unit of analysis.

Results And Discussion

Across all five studies, 1,696 children (54% Black, 46% White; 51% female) were measured. Table 2 shows the percent and number of children in each BMI category overall and by ethnicity and gender. Across all five studies, 1.2% of children were underweight, 58.3% were at a healthy weight, 16.6% were at risk of overweight, and 23.9% were overweight. The mean BMI-for-age percentile was 70.1%. It was greater for Black (72.0%) than White (67.8%) children (p=0.02, analysis of covariance); however, given the large sample of 1,696 and p-value 0.02, the means actually are quite close. What is of more interest is that the mean BMI-for-age percentiles are considerably larger than the 50% that would be expected if the population from which we sampled was like the one on which the norms were based. None of the two- or three-way interactions in the analyses of covariance were significant (p’s>0.27).

A higher percentage of Black (26.5%), as compared to White (21.0%) children, were overweight (p=0.01, chi-square). No differences were found in the prevalence of overweight by gender (p=0.29, chi-square; females=22.9%, males=25.1%) or time (p=0.56, chi-square; Time 1=24.6%, Time 2=22.6%, Time 3=23.0%, Time 3.5=27.7%). No differences were found in the prevalence of at risk of overweight by ethnicity (p=0.23, chi-square; Black=17.6%, White=15.4%), gender (p=0.04, chi-square [not significant after adjustment for multiple comparisons]; females=18.4%, males=14.7%), or time (p=0.37, chi-square; Time 1=15.2%, Time 2=16.2%, Time 3=17.8%, Time 3.5=16.8%).

The rate of overweight among this sample of Fourth-Grade children in Richmond County, GA, at 23.9% between Spring 2000 and Fall 2002, is higher than the national rate of 16%, reported in NHANES 1999-02 for youth ages six to 19 years (CDC, 2006a). Data collected elsewhere in Georgia also indicate greater percentages of overweight children than the national rate. For example, the Georgia Center on Obesity and Related Disorders conducted the Washington- Wilkes Health Study in late Spring 2002 (Davis et al., 2005). ( Washington-Wilkes, GA, is approximately 50 miles west of where the five studies for the current article were conducted.) Results from the Washington-Wilkes Health Study, with 211 youth in Grades Two, Four, Six, Eight, Ten, and Eleven, showed that 20.9% were at risk of overweight and 27.5% were overweight (Davis et al., 2005). For the 38 Fourth-Grade children measured, 61% were at risk of overweight. Thus, the prevalence of youth who were overweight or at risk of overweight in a nearby rural community was slightly higher than that in our sample. However, the participation rate in the Washington-Wilkes study was 37% (Davis et al., 2005), while participation rates in our five studies were higher and ranged from 57% to 73% (Table 1).

The findings in these studies, that more Black than White children were overweight, is similar to findings of Kimm et al. (2002) and Hedley et al. (2004), but different from Melnik et al. (1998), Ogden et al. (2002), and Davis et al. (2005). Among 2,379 females enrolled at nine to ten years old in 1987-88 in the National Heart, Lung, and Blood Institute Growth and Health Study, Kimm et al. (2002) reported that the prevalence of obesity (defined as greater than 95th percentile for BMI-for-age from CDC) at age nine was 17.7% for Black females and 7.7% for White females, and at age ten was 18.5% for Black females and 9.4% for White females. For 2,214 children ages six- through 11 years old in NHANES 1999-02, Hedley et al. (2004) reported that the prevalence of overweight (defined as greater than 95th percentile for BMI-for-age based on CDC growth charts) among non-Hispanic Whites (13.5%) was less than that of non-Hispanic Blacks (19.8%) and Mexican Americans (21.8%).

For 692 Second-Grade and 704 Fifth-Grade children in New York City during the 1989-90 school year, Melnik et al. (1998) found that, for Fifth Grade children, more Hispanics were greater than 85th percentile (40.2%) compared to White (32.0%) and Black (29.5%) non- Hispanics, and more Hispanics were greater than 95th percentile (23.9%) as compared to White (19.5%) and Black (16.0%) non-Hispanics,. There were no differences reported between White and Black non-Hispanics. For 1,054 children ages six to 11 years old in 1999-00 NHANES, Ogden et al. (2002) reported that more Mexican-Americans (23.7%) were overweight (defined as greater than or equal to 95th percentile of BMI-for-age based on CDC growth charts) compared to Non-Hispanic Whites (11.8%), but no differences were reported between Whites and non- Hispanic Blacks. For 211 youth in Grades Two through Eleven in Washington-Wilkes, GA, in Spring 2002, Davis et al. (2005) did not find differences in the percent of White and Black youth with BMI greater than or equal to 85th percentile.

The researchers did not find a difference by gender in the prevalence of children who were overweight or at risk of overweight. In the Melnik et al. (1998) study, there was not a gender difference at the 85th percentile cutoff, but more Fifth-Grade males (21.8%) than females (14.6%) were greater than 95th percentile. For children ages six through 11 years old in NHANES 1999-02, Hedley et al. (2004) found that among males, the prevalence of overweight was lower among non-Hispanic Whites (14.0%) and non-Hispanic Blacks (17.0%) compared to Mexican Americans (26.5%); however, among females, the prevalence of overweight was lower among non-Hispanic Whites (13.1%) compared to non-Hispanic Blacks (22.8%). Neither of these differed from Mexican Americans (17.1%). For the Washington-Wilkes Health Study, Davis et al. (2005) did not find a gender difference in the prevalence of overweight or at risk of overweight. For NHANES 1999-2000, Ogden et al. (2002) did not find a gender difference in the prevalence of overweight among children ages six to 11 years old.

Hedley et al. (2004) found no change in the overall prevalence of overweight in youth between NHANES 1999-2000 and NHANES 2001-02. The results of this study are similar, because the authors found no differences in the percent overweight and the percent at risk of overweight from Spring 2000 to Fall 2002. In a study reported by Moore et al. (2002), 121 Black and 132 White children were measured at an initial visit (mean age 8.8 + 2.0 years) and again seven years later. Children were classified as overweight if BMI was greater than 85th percentile, which differs from the definitions used for this article. Moore et al. (2002) found an increase in the prevalence of overweight (30.8% to 40.3%) for all children across seven years. The increase differed by socioeconomic status (SES), but not by ethnicity or gender. Lower-SES youths exhibited the greatest increases in percent overweight (Moore et al., 2002). In a study with approximately 1,800 Second- and Fifth-Grade children in New York state between Spring 1987 and Spring 1988, Wolfe et al. (1994) found that children who tended have a higher BMI were members of low-SES, two-parent households. For our five studies, the authors only had SES information at the school-level ( Table 1), so it was not possible to analyze the data by SES at the individual child level. The authors did not analyze the data by school because the number of schools differed for each time period, and only two of the 14 schools (B and E) were used during all four time periods for the five studies ( Table 1).

There are several limitations to the findings of this research. These five studies were not designed to assess the prevalence of overweight or at risk of overweight. Analyses were performed on cross-sectional, not longitudinal, data. There were not enough Hispanic students to analyze their data. Most (64%) of our sample came from study periods Two and Three (Studies Two through Four). But these limitations are offset by two strengths. First, measurements were obtained by research staff instead of being self-reported by children; studies have shown that youth ages 11 through 20 years under report weight (Fortenberry, 1992; Himes & Story, 1992; Shannon et al., 1991; Tienboon et al., 1992) and over report height (Fortenberry, 1992; Tienboon et al., 1992). Second, inter-rater reliability of measurements was assessed throughout data collection for each of our five studies. Publications of studies by Kimm et al. (2002), Moore et al. (2002), and Hedley et al. (2004) failed to mention assessment of inter-rater reliability. Wolfe et al. (1994) reported inter-rater reliability ranging from 96% to 100%. Melnik et al. (1998) reported that “periodic retraining of the field team… was conducted throughout the course of the data collection phase”; however, inter-rater reliability was not specifically reported.

Conclusions And Applications

The overall rate for overweight of 23.9% in this Georgia sample was 1.5 times higher than the national rate of 16%, and both are higher than the rate set forth by the Healthy People 2010 objective, which is to reduce the proportion of overweight children to no more than 5% (U.S. Department of Health and Human Services, 2000). Our findings of high prevalence of overweight and at risk of overweight in youth are of concern, considering that Richmond County has cardiovascular disease death rates that are in the top 25 of 159 counties in Georgia (Rowe et al., 1999) and overweight youth are at increased risk for future cardiovascular disease (Georgia Department of Human Resources, 2000).

In 2003, the American Academy of Pediatrics Committee on Nutrition issued a policy statement concerning the prevention of pediatric overweight and obesity. It recommended that: 1) children’s BMI be calculated and plotted once a year; 2) parents and caregivers promote healthy eating patterns by offering nutritious foods; and 3) policy makers from local, state, and national organizations and schools be enlisted to support healthful lifestyles for all children (Committee on Nutrition, 2003).

Meanwhile, in Arkansas, Act 1220 mandated schools to weigh and measure students and to provide parents with an annual BMI percentile (Arkansas General Assembly, 2003). According to a study by Chomitz et al. (2003) that evaluated such a school-based health report card on family awareness of and concerns about children’s weight status, parents of overweight children who received health and fitness reports were more aware of their child’s weight status than those who did not receive any information. In addition, these parents were more likely to report initiating or intending to initiate activities related to weight control. Examples of strategies being implemented in many schools, as well as types of legislation under consideration by several states to improve children’s nutrition, are provided, along with the dietary recommendations for children and adolescents, by the American Heart Association and the American Academy of Pediatrics (Gidding et al., 2005).

In a joint position statement, the American Dietetic Association, the Society for Nutrition Education, and the American School Food Service Association [now the School Nutrition Association] (2003) asserted that:

“Comprehensive nutrition services must be provided to all of the nation’s preschool through grade twelve students. These nutrition services shall be integrated with a coordinated, comprehensive, school health program and implemented through a school nutrition policy. The policy should link comprehensive, sequential nutrition education; access to and promotion of child nutrition programs providing nutritious meals and snacks in the school environment; and family, community, and health services’ partnerships supporting positive health outcomes for all children.”

The Child Nutrition and WIC Reauthorization Act of 2004 required local education agencies that sponsor school meal programs to establish local school wellness policies addressing childhood obesity by the beginning of the 2006-07 school year (U.S. Congress, 2004). Results from focus groups conducted with school foodservice directors in Pennsylvania in June 2005, found that, although the majority anticipated they would play a lead role in the development of the required local wellness policies, many expressed varying degrees of comfort with this role (McDonnell et al., 2006).

The position of the American Dietetic Association (2006) is that “schools and the community have a shared responsibility to provide all students with access to high-quality foods and school- based nutrition services as an integral part of the total education program. Educational goals, including the nutrition goals of the National School Lunch Program and School Breakfast Program, should be supported and extended through school district wellness policies that create overall school environments that promote access to healthful school meals and physical activity and provide learning experiences that enable students to develop lifelong healthful eating habits.”

Large and extra-large portion sizes have been implicated as an environmental contribution to the obesity epidemic (Hill & Peters, 1998). School meal programs provide age-appropriate serving sizes to the millions of children who participate in school breakfast, school lunch, and/or school snacks on a daily basis. Thus, school foodservice directors may want to use serving sizes provided for school meals and snacks to help educate children, teachers, and parents about age-appropriate serving sizes.

Because food assistance programs serve many individuals who are obese, there is concern that participation in food assistance programs may precipitate the growing problems of obesity (Hofferth & Curtain, 2005; U.S. Department of Agriculture, 2005). However, an expert panel convened in March 2004 to identify scientific evidence about this concern found no published research providing evidence of a consistent relationship between childhood obesity and participation in the School Breakfast Program and National School Lunch Program (U.S. Department of Agriculture, 2005). Research to date about this relationship has been limited by:

1) categorizing children as “fed” or “non-fed,” based on administrative records of children’s participation in school lunch during the previous school year (Paige, 1971); 2) parent reports of children’s average weekly participation in school breakfast and school lunch (Vermeersch et al., 1984); 3) parent “yes” or “no” reports of children’s participation in school lunch (Melnik et al., 1998; Wolfe et al., 1994); and 4) parent “yes” or “no” reports of children’s participation in school breakfast and school lunch within the previous year (Hofferth & Curtain, 2005; Jones et al., 2003). To the authors’ knowledge, no study has investigated the relationship between childhood obesity and daily participation in school meals (when daily participation is a continuous variable). Although weight and height measurements were included as secondary aims in our five studies, unfortunately, parental permission was not obtained to permit the school nutrition program to release information about children’s daily participation in school meals.

Although it is encouraging that more schools are obtaining annual weight and height measurements, certain methodological aspects deserve consideration. First, the time of day of measurements should be standardized from one year to the next when tracking individual children’s annual BMIs. Second, measurers should be properly trained, and inter-rater reliability should be performed throughout data collection to ascertain that measurements do not vary based on the individual taking the measurements. Height can be especially challenging to measure, and differences of one inch can change a child’s BMI category. (For example, using the “Child and Teen BMI Calculator” [CDC, 2006b], a female who weighs 75 pounds at 9 years old is categorized as having a healthy weight if her measured height is 53 inches; but she is at risk of overweight if her measured height is 52 inches.) Third, obtaining measurements in a private location at school is important to help relieve concerns that some children may have about other children knowing their weights.

Schools provide an excellent venue for measuring children’s weight and height annually to determine BMI-for-age percentiles. This information can be used to track changes over time, which can help schools determine whether the implemented wellness policies are effective and provide ongoing information about the health status of the school population (Crawford et al., n.d.). Future research could link children’s annual BMI-for-age percentiles to computerized administrative records of children’s daily participation in school meals from one year to the next to increase knowledge of a relationship between childhood obesity and daily participation in school meals.

Acknowledgments

The authors express appreciation to the children, faculty, and staff of Blythe, Glenn Hills, Goshen, Gracewood, Hephzibah, Lake Forest Hills, McBean, Monte Sano, National Hills, Rollins, Southside, Terrace Manor, Willis Foreman, and Windsor Spring Elementary Schools and to the Richmond County Board of Education in Georgia for allowing data to be collected. The authors also thank Michelle Baglio, RD; Francesca Frye, MBA, RD; Candace Kopec, PDt; and Nicole Shaffer, RD, for help with data collection. Finally, the authors thank Julie Royer, MSPH, for her help with revising the manuscript. Funding was from R01 HL63189 from the National Heart, Lung, and Blood Institute of the National Institutes of Health, and a State of Georgia Biomedical Initiative grant to the Georgia Center on Obesity and Related Disorders at the Medical College of Georgia. Suzanne D. Baxter was Principal Investigator of both grants.